JustAnswer's Journey to Chat-First Support

JustAnswer implemented chatbot technology in its Help Center to improve self-service options. Testing revealed higher resolution rates for chat compared to email and phone support, leading them to adopt a chat-focused approach that saved $119,000 annually.

Problem/Opportunity: JustAnswer's chatbot technology has evolved significantly over the years. We aim to leverage chat functionality to enhance self-help options for customers in the Help Center. This presents a valuable opportunity to reduce customer service costs through AI-powered support across multiple channels.

Hypotheses: Chat support shows higher self-help resolution rates compared to email and phone channels. Therefore, directing customers toward chat interactions should decrease overall support tickets.

The goal of the A/B test: To increase chat-based customer service interactions while maintaining overall customer service engagement levels.

Impact: The initial solution saved $119K/year.

Role: UX Manager and Designer (work with PM, UX Designer, Content Strategist, Data Analyst, stakeholders, Engineering, and Conversation Designer)

Insights | Opportunity | Solutions | Challenges | Test Results & Optimization

Insights

Customer incidents by source and issue distribution before the test launch:

  • Sources: Chat 21%, Phone 40%, Email 39%

  • Top 2 issues: Membership 30%, Cost & Payments 23%

  • ~5M in agent costs per year

  • Current Resolution Rate (incidents resolved by chatbot or IVR) = 16%, which saves $800K+

  • Resolution Rate by Chat only = 27%, which is higher than any other sources

Opportunity

Having more customers using chat as a primary source of Customer Service contact on the Help Center, we'll be able to spend more consolidated efforts on improving and optimizing the chatbot.

Before (control version)

Solutions

  • Emphasize the chat option and have phone/email still available but de-emphasized.

  • Adding benefits for using the chat to solve issues.

  • Provide trust content to give confidence to the customers.

User Testing Summary

  • The overall design and layout inspired trust and positivity.

  • Chat option has become a natural way to contact Customer Service.

  • The human picture on the site conveyed that there was a human behind the Help Center.

Challenges

  • Backend infrastructure can only run one Customer Service Bot at a time; therefore, there are only two tests per quarter.

  • Past tests had a negative financial impact and drop in contacts to Customer Service overall

Test Results

  • The primary metric (+15% chat contacts per customer on Help Center) is significantly up

  • The current implementation did not result in LTV drop in variation

  • Customers in chat-only variation still manage to find ways to contact CS via phone and email

Normalizing the test may result in savings of $119k/year.  

Optimization

  • With the new layout normalized, the focus is to improve dialogue flow and AI script.

  • 1st optimization: Customer Service Bot Opt-in (Chat Bot Pearl will now give the customer a choice on whether to be transferred to reduce human agent cost)